June 21, 2013
Here at the close of the International Supercomputing Conference and the announcement of the new Top 500 rankings, we wanted to collect and create a more palatable set of visualizations to put this year's list in more context.
The following chart, as taken from the Top 500, shows the top ten with performance detail--we'll move on to more specific breakdowns in a moment, but it does provide a handy way to see the performance differences between multiple machines.
On that note, take a look at the difference between the top Tianhe-2 system and the second in line--Titan at ORNL as well as the line of performance developments leading up to both of these systems in particular.
One of the major themes during this show--and to some extent the last few Top 500 rankings--is the role that accelerators and coprocessors are playing in supercomputing. IDC, Intersect360 Research and the Top 500 founders agreed this week that within a few years, barring any unexpected technology developments, coprocessors and accelerators will dominate the entirety of the top ten.
As the marketplace changes, however, it's not just GPUs leading the pack. This was arguably Intel's show with the Tianhe-2 win and a string of announcements from other vendors lining up to support the future of Xeon Phi.
Ultra high performance, coprocessors and accelerators or not, means massive power consumption. The list below, this one taken from the Top 500, shows a broad mix of architectures that have proven the most efficient--these include accelerated systems fed by AMD, Intel and IBM...
Switching gears, we took a look at the list to get a sense of where the academic/research users split from the industrial HPC user base. A quick comparison showed that the use areas has remained steady for the last couple of years--but also shows that top systems in industry take a much wider piece of the pie, even if it tends to be the big national labs' academic and research machines that steal the attention.
Of those systems, there are some clear demarcations along vendor lines and, as we will see in a moment--along processor vendors as well. Again, not a huge change over the last year or two on the system vendor front but a steady presence on the part of HP and IBM.
Processor Vendor Share
Intel's share of the processor market is worth a closer look in terms of products. While Westmere will likely dip between now and November during the string of upgrades, it's worth breaking down the Xeon share for a clearer view.
Operating System Share
For more visualizations from the Top 500, take a look at the following presentation:
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?